Abstract

Organisations and individuals release source code on the Web to improve their software by attracting peers in the strategic move of “opensourcing” that has created thousands of open source projects (e.g., Eclipse-IBM, Thunderbird-Mozilla and Linux-Torvalds). Nevertheless, most of these projects fail to attract people and never become active. To minimize this problem, we developed a theoretical model around a crucial construct (attractiveness) to open source projects, proposing its causes (project characteristics), indicators (e.g., number of members) and consequences (levels of activeness, efficiency, likelihood of task completion, time for task completion and software quality). We tested this model empirically using 3 samples of over 4600 projects each in a multi-sample SEM analysis. The results confirm the central role that attractiveness plays to guarantee an active and efficient community of software development, shedding new light on whether more developers increase software quality by finding and fixing more bugs and providing upgrades. They also clarify the actual causal structure involving Web page visits, downloads and members, which can be easily mistaken. Moreover, the results can provide useful insights to strategists as we discuss the impacts of license restrictiveness, software development status, type of project and intended audience on attractiveness and its consequences.

Share

COinS